Needed - help from a "stat" freak - - -

[KSFW&L]

I have a friend who is new to online poker, but is - IMO - a very solid player.

She started playing two weeks ago, but has logged a huge number of hours - all at 3-6.

Her record is as follows...

Total hands played: approximately 10,000

Total hours played: ~ 100

- I'm estimating that this would equate to roughly 400 hours of "live" play. Online can and often does yield 100+ hands per hour; 3-6 in a cardroom seldom produces more than 25 or 26 HPH.

She is ahead just over $2,000 - this is not a staggering amount when you break it down to $/hour; her $20/hour online win rate would only equate to ~ $5/hour in a live game.

1. What is the likelyhood that she is at least a $10/hour online winning player who has run very hot ?

- My guestimate is that there is a 75%+ chance that she does play well enough to win $10+/hour in the long run.

2. What is the likelyhood that she is at least a $15/hour online winner who has run a little hot ?

- Here I'm guessing there is ~ a 50/50 chance that she is THIS good.

3. Is there a "statistically significant" possibility that she is a losing player who has been getting hit with the deck since she began play ?

- Having watched her play in live games I would be very suprised if the answer to #3 was yes, but I would love to hear from someone better qualified to answer than myself.

If there is a way for her standard deviation to be calulated accurately given this relatively small sample size I would be eternally grateful.

Thanks for any and all replies.

- L -

[BruceZ]

The questions you ask can only be answered once you compute her estimated standard deviation. To compute that, you need her results logged for each session, and the duration of each session. The formula is available in the essay section, or if all sessions are the same length you can use my simple method below. It should be fairly accurate after this many hours, and the accuracy can be quantified if you know the number of sessions. The estimation of her hourly rate or EV by simply dividing her winnings by the number of hours played will not be very accurate at all, and a large range of EVs is possible after this amount of time, however:

My guestimate is that there is a 75%+ chance that she does play well enough to win $10+/hour in the long run.

This is an example where you could use Baysian analysis to estimate her EV more accurately. Baysian analysis takes into account your predisposition for where her EV should lie.